一种基于三阶亮度校正的平面视频转立体视频快速算法

本文提出了一种新的平面视频转立体视频的快速算法.这种算法能够实时的将平面视频转换成立体视频,并能在三维显示设备上呈现出逼真的立体效果.首先将原始平面视频中的图像按照高斯分布进行立体变换,然后将视频中的图像序列生成加权平均亮度图,并将亮度分为3个等级,分别对这3个等级区域中的图像进行立体校正,最终得到完整的立体视频.我们的方法替代了传统方法中,生成深度图像的步骤,从而大大的提升了运算的速度,能够在原始平面视频的实时播放过程中,直接输出带有立体效果的画面.

An effective algorithm for monocular video to stereoscopic video transformation based on three-way Iuminance correction

This paper presents a new effective algorithm for monocular video stereoscopically transformation.With this algorithm,the monocular video can be transformed into stereoscopic format in nearly real time,and the output stream can be shown with lifelike three-dimensional effect on any supported display device.The core idea of this algorithm is to extract images from original monocular video,transform the images into stereoscopic ones according to Gaussian distribution,then build a three-level weighted average brightness map from the generated stereoscopic image sequences,correct the image regions respectively in all three level,and finally compose the complete three-dimensional video.After replacing the traditional time-consuming depth image generation algorithm with this one,the transformation performance obtains significantly improvement.Now the images with three-dimensional stereoscopic effect can be shown in real time during the original monocular video live broadcasts.